okay so i keep noticing that ai copilots make it feel like i’m moving faster, but not always like i understand the code any better. that gap is kind of weird — especially when the suggestion is close enough to ship but not close enough to trust.
has anyone else run into that? like, the autocomplete is impressive, but the confidence part still seems to come from the boring manual review.
yeah, that’s been my experience too. it’ll spit out something that compiles, but the part that bites you is usually some tiny assumption — like it handled null fine but quietly got the ordering wrong, or it “helpfully” caught the wrong exception.
for me the confidence comes back only after i read the diff and run the tests, which is kind of annoying because that’s the slow part anyway. do you ever make it explain the change in plain english before you accept it, or does that just feel like extra ceremony?
:: Copyright KIRUPA 2024 //--